I am trying to create an evaluation function for a general game player based on the research from this article An Automatically-Generated Evaluation Function in General Game Playing. I can't understand the 2nd phase of this process.
The evaluation function in this article consists of 4 phases.
Features identification (generalization and specialization)
More specifically, in the 2nd phase, we will make 100 random short game sequences from a starting position (let's suppose we play tic-tac-toe) and 10 rollouts from every sequence. Then we calculate the variance and stability on each feature based on their values, average values, etc.
I would like to mention I don't even understand what the score of the feature supposed to be? How am I going to find it, and then how I will find the variance of the feature?